The Influence of Country of Origin in the Process of Party Identification Acquisition

Abstract

Thousands of Latin Americans migrate to the United States every year. This article seeks to understand how immigrants’ premigration political experiences influence the acquisition of party identification upon arrival in the United States. This research proposes that premigration political experiences influence the acquisition of party identification among Latino immigrants in the United States. Utilizing data from the 2016 Collaborative Multiracial Post-Election Survey (CMPS) and Proyecto Élites Parlamentarias Latinoamericanas (Latin American Parliamentary Elites Project), this paper analyzes how the ideology of the government in power in the immigrants’ country of origin influences party identification among Latino immigrants in the United States. Employing multinomial regression analysis, I demonstrate that the ideology of governments in power in Latin American countries when Latinos migrate influences the party identification of those immigrants in the United States. The results of this study contribute to the conversations on premigration experiences and challenge the applicability of classical theories of party identification for immigrants.

Introduction

Immigrant party acquisition and Latino party identification have been generally analyzed through classical theoretical frameworks that mainly consider determinants occurring in the United States. In the last decades, much of the analysis of Latino immigrant party acquisition has involved postmigration factors such as mobilization efforts (Garcia and De La Garza 1985), economic situation, foreign policy issues (Cain, Kiewiet, and Uhlaner 1991), or English language skills (Wong 2000). Although these factors have contributed to understanding Latino political behavior, they do not tell the whole story. Immigrants come with premigration political experiences and knowledge, which influence their political behavior after arrival.

Following this line of thought, some scholars have paid attention to analyzing premigration experiences of Latino immigrants. More recent research posits that political knowledge and experiences before migrating to the United States influence political behavior. Wals (2011) argues that immigrants come from their countries of origin to the United States with “political suitcases,” referring to a background of attitudes and experiences. To better understand postmigration political behavior, Wals suggests we must “peek inside these suitcases” (p. 601). For example, Ishiyama and Silva (2020) examined the impact of premigration experiences and found that premigratory experiences with ethnic violence shape political opinion in the United States. Antoine Bilodeau and Ian McAllister (2005) found that politics in the home country influences immigrants’ political behavior in the host country. Similarly, Wals and Rudolph (2019) found that the degree of democracy in countries of origin during premigratory socialization can shape postmigration political trust in the United States.

National origin is perhaps the most studied premigration factor in understanding Latino immigrants’ political behavior (Pachon and DeSipio 1994). Researchers generally agree that Cubans are often Republican conservatives, while Mexicans and Puerto Ricans tend to be Democrats (Alvarez and Bedolla 2003). However, while scholars have studied how national origin influences the political behavior and party identification of immigrants, our understanding of the influence of political backgrounds within countries of origin could be improved.

To this end, this paper sheds light on how premigration political experiences in countries of origin influence how immigrants navigate American politics. This article addresses the following question: Is party identification among Latino immigrants influenced by the ideology of the government in their home countries at the time of migration to the United States? I argue that when Latinos migrate, the government’s ideology in the country of origin influences party identification once the immigrant is in the United States, even when controlling for potential covariates. Having migrated from a country with either a left- or right-wing government significantly influences the partisanship of Latino immigrants.

Latinos Before and After Migration

The principal theories of migration studies are based on push and pull factors that motivate people to migrate (Lee 1966). Push and pull factors can be classified as economic, demographic, sociocultural, political, and miscellaneous factors (Kumar and Sidhu 2005). Rumbaut (1994) demonstrated that economic considerations carry significant weight in the decision to migrate.

Push factors are all those reasons that persuade a person to leave their country of origin and move to another one. Some push factors include unemployment, low wages, poverty, natural disasters, and political threats such as dictatorships, armed conflict, or civil wars. Push and pull factors can explain Latin American migration to the United States (Rendon and Cuecuecha 2010). To a greater or lesser extent, these factors affect or have affected all Latin American countries and have motivated people from these countries to migrate to the United States in search of opportunities and a better quality of life (Bedolla 2009).

Politics can be considered a push factor. For example, during the 1960s and 1990s, right-wing governments in Argentina, Brazil, Bolivia, Chile, Paraguay, and Uruguay persecuted, tortured, and murdered thousands of leftist or socialist people or anyone suspected of being left-wing or socialist. As a result, thousands of people emigrated to other countries. Similarly, thousands of Cubans have emigrated from Cuba, and first-generation Cubans in the United States offer a historical example of immigrants whose political experiences with a left-wing government have influenced their party identification in the United States. Cuban immigrants tend to see the Cuban revolution as a negative outcome (Uhlaner and Garcia 1998), suggesting that waves of Cuban immigrants are composed mainly of those who opposed the Cuban revolution before leaving Cuba. Similarly, Mexicans in New York are more likely to support Mexican opposition parties than Mexicans in Mexico (Lawson 2003). Once in the United States, Cubans identify mostly with the Republican Party. Research has found that Cuban immigrants are more likely to be Republicans due to their attitudes toward communism and the 1959 Cuban revolution (Calderon 2019; Uhlaner and Garcia 1998). I argue that this influence of government ideologies in countries of origin on first-generation immigrants is not exclusive to Cubans or Mexicans and can be found in Latino immigrants from different countries.

Immigrants arrive with experiences acquired in their home country and with prior political education, which influences their political assimilation in their new place (Finifter and Finifter 1989; Guarnizo, Chaudhary, and Sørensen 2019; Wals 2011; Wals and Rudolph 2019). Immigrants can carry their ideology from their home country to the United States (Wals 2013). Once in the United States, this “political suitcase,” in Wals’s (2011) terms, can act as a lens to see and interpret American politics, including party identification.

I investigate the relationship between the government ideology of first-generation Latino immigrants when they left their home country and their U.S. party identification. I hypothesize that first-generation immigrants develop a partisanship when they migrate that contradicts the government ideology in their home country. Specifically, I expect that first-generation Latino immigrants from a country with a left-wing government are more likely to identify with Republicans. By contrast, those first-generation Latino immigrants who move from a country with a right-wing government are more likely to identify with Democrats. This research offers a statistical analysis that supports my hypothesis and points to a different theory of party acquisition for U.S.-born versus foreign-born people.

Partisanship in the United States

Scholars have conducted extensive research on the factors that influence party identification (Berelson, Lazarsfeld, and McPhee 1986; Campbell et al. 1960; Downs 1957; Fiorina 1981; Green and Palmquist 1990; Markus and Converse 1979). According to the socialization model of partisanship (Campbell et al. 1960), party identification is influenced by affective attachments to a social group. These attachments are learned early in life and remain stable over time (Abramson and Ostrom 1991; Green and Palmquist 1990; Miller 1991). Converse (1969) suggested that age reinforces party identification, but it should apply only when the “passage of years in chronological age” equates to the length of exposure to the political environment of the United States (p. 143). In first-generation immigrants, this affirmation is not valid.

The revisionist model consists of a more unstable vision of partisanship than the socialization model, in which identification is more malleable (Fiorina 1981; Franklin 1984; Franklin and Jackson 1983; Markus and Converse 1979; McKuen, Erikson, and Stimson 1989). According to this model, political events can change party identification, and people can change their political affiliation repeatedly over a lifetime. Fiorina (1981, p. 90) stated that “when a citizen first attains political awareness, socialization influences may dominate party identification. But as time passes, as the citizen experiences politics, party identification comes more and more to reflect the events that transpire in the world.”

Classical theories of party identification assume that political preferences are based on experiences in the United States. Consequently, scholars have focused on contextual variables primarily related to experiences in the United States after an immigrant’s arrival (Wals 2011). Many political scientists have started with these theories as a basis for describing Latino and immigrant party identification. However, the literature of the past decades indicates that premigration experiences influence political behavior in the host country (Black, Niemi, and Powell 1987; Finifter and Finifter 1989; Wals 2011; Wals and Rudolph 2019). The importance of this research lies precisely in the rationale that Latino immigrants’ premigration political experiences are essential in shaping political preferences. Therefore, this research considers these experiences in order to better understand immigrant partisanship acquisition.

Applicability of Conventional Models to Latino Immigrants

Many scholars have suggested that conventional models of party acquisition are less applicable to immigrants (Alvarez and Bedolla 2003; Sears, Danbold, and Zavala 2016; Uhlaner and Garcia 1998, 2005) or Latinos (Cain, Kiewiet, and Uhlaner 1991) than to native-born (White) Americans, and these models may not fit the political experiences of minority or immigrant groups. For example, age is a critical variable for conventional party identification models that measure exposure to U.S. politics. Some studies suggest that people become more conservative (Uhlaner and Garcia 1998) and Republican (Ocampo and Ocampo 2020) as they age. For immigrants, however, age does not translate into cumulative exposure to the U.S. system. While for the U.S.-born, age is an indicator of experience and exposure to the U.S. political system, it is not so for immigrants because their experiences of the U.S. political system started later, when they entered the United States. Upon arrival, an immigrant may be considered “born into” the U.S. political system. By 2015, 34.4% of Latinos in the United States were immigrants (Flores 2017), and many Latinos had been socialized, at least partially, outside the United States (Abrajano and Alvarez 2010). This forces the separation of chronological age for Latino immigrants from experience and exposure to the U.S. political system. Previous scholarship has found that for Latinos and Asians, the length of time in the United States is associated with party identification (Wong 2000) and contributes to an increased likelihood of Democratic self-identification (Barreto and Woods 2005; Cain, Kiewiet, and Uhlaner 1991).

Parental socialization is another characteristic that affects the application of conventional theories to Latino immigrants (Uhlaner and Garcia 1998). Parental socialization does not play the same role for immigrants because traditional models of party acquisition argue that immigrants (or their parents) do not receive significant socialization in the U.S. context. Uhlaner and Garcia (1998) and Hajnal and Lee (2011) argue that the migration experience, parental socialization, and engagement with the U.S. political system affect the applicability of traditional theories for Latino immigrants, especially rational choice models, because they emphasize the extent of information available for party evaluation (Downs 1957). In rational choice models, people need to have considerable information to translate their experiences into a decision regarding which political party is more aligned with their interests. Since first-generation Latino immigrants do not have U.S. parental socialization, they use their political experiences and the knowledge acquired in their countries of origin to inform that decision (Hajnal and Lee 2011).

Linked fate is another factor that may influence party identification among Latinos. Linked fate is a commonly used form of group consciousness among African Americans (Dawson 1994) who perceive that the problems affecting their group also affect themselves and that resolving the issue affecting the group will help their individual lives. This contrasts with rational choice assumptions of self-interest, according to Anthony Downs (1957). Research has found that linked fate operates similarly for Latinos (Sanchez and Masuoka 2010) and can influence party identification among Latinos because they could think about their interests as a minority group.

Latino Party Identification and Political Behavior

Scholars initially looked at national origin and standard socio-economic status (SES) to explain Latino party identification. Scholars have found, for example, that factors such as national origin, income, education, unemployment, immigration, age, religion, sex, and bilingualism influence Latino party acquisition (Uhlaner and Garcia 1998). Mexicans and Puerto Ricans in the United States tend to be Democrats, while Cubans tend to be Republicans. In addition, there are differences in preferences and patterns between immigrants and native-born citizens (Dancygier and Saunders 2006). These differences are not limited to immigrants versus native-born: Among people who think of themselves as from the same country, there are also significant differences between first-generation immigrants and native-born citizens (Dancygier and Saunders 2006). For example, 58.82% of the foreign-born Mexicans who answered the 2016 Collaborative Multiracial Post-Election Survey (CMPS) (Barreto and Lorrie Frasure-Yokley 2017) are Democrats. By contrast, only 46.58% of their U.S.-born counterparts considered themselves Democrats. This difference exceeds 12% when accounting for weighted data. With different directions and dimensions, these differences in patterns are evident among Latino nationalities in the United States.

Historians and political scientists have argued that Cubans, Puerto Ricans, and Mexicans living in the United States have been politically influenced by their historical circumstances (Calderon 2019). First-wave Cubans offer an illustrative example. In the United States, they identify mostly as members of the Republican Party. Research has found that this is likely a result of their attitudes toward communism and the 1959 Cuban revolution that installed a left-wing government (Calderon 2019; Uhlaner and Garcia 1998). Moreno (1997) suggests that Cubans lean toward the Republican Party because of institutional and political issues such as Latin American communism and the United States’ anti-Castro policies. Conversely, de la Garza (2004) posits that immigrants are focused on U.S. politics rather than politics related to their country of origin. However, there is a generational difference among Cubans in the United States. Second-generation Cubans, who do not have premigration experiences in Cuba, do not identify as Republicans to the same extent as first-generation Cubans.

U.S. scholarship on Latino partisanship has advanced beyond its initial consideration of national origin and the standard SES variables to include contextual variables. This work has primarily relied on political preferences and experiences after migrating to the United States. For example, Alvarez and Bedolla (2003) found that policy preferences on issues of partisan political interest, such as abortion, undocumented immigration, or gun control, explain Latino partisanship better than standard SES variables. ¹ Later, they found that although social, political, and economic issues influence Latino party acquisition, the first two can better explain party identification (Alvarez and Bedolla 2003).

Migration has become among the most studied policy issues in recent years, especially regarding its effects on Latino political behavior. In 2010, approximately 81% of both Latino immigrants and second- and third-generation U.S.-born Latino voters in Arizona vehemently opposed the Immigration Law SB1070 (Barreto and Segura 2010), which defines a misdemeanor crime for immigrants not carrying a registration document. They also found that more than 80% of Latinos were concerned that Latinos who are legal immigrants or U.S. citizens would be stopped or questioned by police. The fact that many Latinos are immigrants (Flores 2017), have close ties with undocumented or deported people (Street, Zepeda-Millán, and Jones-Correa 2015), or are themselves undocumented influences such concerns. Researchers found that survey participants strongly supported the legalization of undocumented immigrants. Immigrant policies have a direct impact not only on political behavior but also on Latino civic engagement with public services. Undocumented Latinos tend not to seek services that are broadly available to the public, such as health care (Pedraza and Osorio 2017; Toomey et al. 2014), calling the police if necessary (Ammar et al. 2005; Pedraza and Osorio 2017), or meeting with their children’s teachers (Pedraza and Osorio 2017).

Mass deportation policies are crucial for Latinos, potentially reshaping their partisanship. Street, Zepeda-Millán, and Jones-Correa (2015) found that Latinos’ support for the Democratic Party substantially decreased after learning that more deportations occurred under President Obama than under his predecessor, President G. W. Bush. Anti-immigrant and anti-Latino measures in the 2000s pushed Latinos to the Democratic Party (Barreto and Woods 2005; Bowler, Nicholson, and Segura 2006). Saavedra Cisneros (2017) proposed a theory to understand immigration partisanship in terms of Latino party identification: Immigration issues can define Latino partisanship, so Republican anti-immigration rhetoric and policies “push [them] away” from the Republican Party (p. 2).

During the spring of 2006, millions of Latinos protested in the streets against the threat of House Bill 4437 (HR 4437), which would have made anti-immigrant laws more aggressive. Recent literature about Latino mass mobilization has developed exclusively on the back of those protests (Wallace and Zepeda-Millán 2020). Some determinants for participation in these 2006 protests among Latinos included, at the individual level, age, gender, education, income, religion, belonging to an organization, and U.S. citizenship. At the geographic level, the most important predictor was the total Latino population in a city (Martinez 2008). More than a decade later, Latinos mobilized protests several times for reasons related to anti-Latino and anti-immigration policies during the Trump administration. The racial threat that President Trump represented to Latinos and immigrants in 2016 influenced Latino mobilization. Attacks by Donald Trump during his 2016 campaign focused mainly on Mexicans and immigrants. However, Latino pan-ethnicity made Latinos of other national groups feel attacked, leading them to protest and participate in rallies (Gutierrez et al. 2019). Access to information is important in Latino partisanship. Overall, Latino voters have a reasonably good level of knowledge about policy issues and candidates’ political positions (Nicholson, Pantoja, and Segura 2006). Documented Latino immigrants acquire more political knowledge the longer they live in the United States (Brown and Bean 2016). It takes years for Latino immigrants to learn about U.S. politics (Fraga et al. 2011; Hajnal and Lee 2011). As political information in the United States is more available in English than in Spanish, having English skills reduces information barriers to the U.S. political system. English proficiency is strongly related to Latino voter participation (DeSipio 1996; Tam Cho 1999), and the language in which an interview is conducted is associated with political opinion (Lee and Pérez 2014). In addition, Latinos have a wide diversity of religions (Kosmin and Keysar 1995), which can be an important factor in determining party identification. Latino Catholics are more likely to be Democrats, while Latino Protestants are more likely to be Republicans (Kosmin and Keysar 1995; Uhlaner and Garcia 1998).

Theory and Hypothesis

Migrants leave their countries due to economic, political, and ideological factors. However, ideology can influence perceptions about the economy and politics, which could drive decisions and behaviors, including the decision to migrate and the process of party identification in a new country. They arrive in the host country with a background of political ideas (Wals 2011), which are likely opposite to those of the government that was in power when they left their countries (Lawson 2003). Once in the United States, and in the absence of U.S. parental socialization and engagement with the American political system, immigrants apply their background political ideas to their interpretation of and identification with political parties. Finifter and Finifter (1989) demonstrated that immigrants use political references from their home countries when acquiring party identification in the host country. Immigrants import their ideology to the United States in terms of intensity and directionality (Wals 2013). Ideologically, they also align their process of party identification to their likely opposition to the government in their home country before they migrated. Consequently, immigrants who leave a country that has a left-wing government will likely import a “conservative political suitcase,” resulting in eventual Republican self-identification once in the United States. By contrast, those who left a country that has a right-wing government will likely arrive with a “liberal political suitcase” and eventually self-identify as Democrat.

Immigrants who entered the United States at an early age, when they had limited political experience and likely little knowledge about politics or policies in their home countries, followed parental socialization relating to the politics of their home country. Second-generation Latinos and Asians do not receive parental partisan transmission that is based on the U.S. context (Carlos 2018), yet these parents, with political and ideological backgrounds imported from their countries of origin, raise that second generation. For instance, second-generation Cubans do not identify as Republicans to the same extent as first-generation Cubans (Krogstad 2014). They are, however, significantly more Republican than non-Cuban Latinos, presumably because, although second-generation Cubans did not live in or leave a country with a left-wing government, right-wing parents who opposed the left-wing communist Cuban revolution raised them. Taking the literature and the proposed theory into account, I develop and test the following hypothesis:

First-generation immigrants will develop party identification opposite to the government ideology in their home country at the time they migrated.

Data, Methods, and Measures

To test this hypothesis, I used individual-level data from CMPS 2016 (Barreto et al. 2018) and contextual-level data about Latin American governments’ ideologies from Proyecto Élites Parlamentarias Latinoamericanas (PELA) (Latin American Parliamentary Elites Project). CMPS 2016 includes approximately 3,000 Latino respondents living in the United States, a much larger sample than other surveys. In addition, the survey includes other racial groups: Whites, Blacks, and Asian Americans. The survey was conducted from December 2016 to mid-February 2017 by Latino Decisions in collaboration with Pacific Market Research. The survey was conducted online, and Latinos responded in English or Spanish. CMPS 2016 gathered information from a nationally representative sample and included a survey design that made the results generalizable to the general population (Barreto et al. 2018). In addition to asking hundreds of political questions, the survey included the standard socio-demographic variables.

CMPS 2016 constructed a sample weight using the 2015 American Community Survey (ACS). This sample weight was meant to balance differences between the CMPS 2016 sample and the adult population in the 2015 ACS in terms of age, gender, education, nativity, ancestry, and voter registration status. As the CMPS 2016 methodological guidelines instruct, I adjusted the analysis to account for complex survey design elements. I incorporated these sample weights into the statistical analysis in order to consider differences between my sample and the general adult Latino population in the United States.

The data for Latin American governments’ ideologies comes from PELA, a project at the University of Salamanca that has interviewed legislators from almost all Latin American countries and governments since 1994. With data from PELA, I measured the ideology of Latin American governments with the president’s position on the left-right scale, according to the evaluation of legislators from different countries per presidential term. Legislators were asked, “On a scale where 1 is left and 10 right, where would you place the following political leaders?” The president is among these political leaders. I used secondary data to develop all analyses in this research and thus did not need Institutional Review Board approval.

From approximately 3,000 Latino respondents in CMPS, the sample for Model 1 contains 325 Latino immigrant respondents. There are several reasons for this reduction. First, only 859 Latino respondents were immigrants. Second, of those 859 respondents, there is PELA data on corresponding presidential ideology for only 370. Finally, I used listwise deletion of missing observations, reducing the sample from 370 to 325. To avoid bias estimates due to listwise deletion, I compared the means for party identification for the 859 Latino immigrants in CMPS with those for whom I have corresponding PELA data about presidents. The mean for party identification for the 859 CMPS respondents is 2.16, while the corresponding mean for the 325 respondents in the final sample is nearly identical, 2.18. Given the similarity in means, I did not expect listwise deletion to produce significant bias estimates.

Deriving conclusions from a model with 325 respondents can be a limitation. To address this concern, I compared the descriptive statistics of the 325 subjects from Model 1 with those of Latino immigrants from the 2016 Cooperative Congressional Elections Study (CCES), adjusting for survey design in both surveys (Ansolabehere and Schaffner 2017). While CCES 2016 is a pre- and post-election survey that includes data on over 64,000 respondents, I focus on a subsample of 1,369 Latino immigrant respondents. The survey includes weights to adjust for the characteristics of the national electorate.

Unfortunately, the CCES 2016 data could not perfectly replicate the study conducted in this paper because CCES did not ask respondents the year they arrived in the United States. Nevertheless, CCES 2016 is helpful in comparing the descriptive statistics of variables that are generally comparable. Overall, my analysis of differences in weighted means is helpful in showing that generally comparable variables in CCES 2016 are similar to those from CMPS 2016 used in Model 1. Income has the highest difference in means (−1.069, weighted), but this is not a concerning difference considering that the variable for income ranges from 1 to 20. In Table 1, I show that the differences in means for comparable variables are minimal.

Table 1. Comparison of variables between the 2016 Collaborative Multiracial Post-Election Survey (CMPS) and the 2016 Cooperative Congressional Elections Study (CCES)

Note: Statistics adjusted for survey design.

Dependent Variable: Party Identification

The primary dependent variable is party identification. CMPS asked respondents, “Generally speaking, do you think of yourself as a Republican, a Democrat, an Independent, or something else?” Respondents selected from among “Republican,” “Democrat,” “Independent,” or “Other party.” I used this question to analyze party identification in order to create a modified categorical measure. I coded Republicans as 1 and, Democrats as 2, and I collapsed Independents and “other party” together as 3.

Researchers often use a seven-point scale for party identification, from strong Democrats to strong Republicans, with Independents in the middle (Miller 1991). For example, if a person chose to answer “Independent” to the first question and then responded either “Democrat” or “Republican” to the third question, they would be considered a weak Democrat or Republican. This scale assumes that Independents are in the middle and are unpredictable and unstable over time. Hajnal and Lee (2006, 2011) have criticized the applicability of this assumption to Latinos. They criticized the application of this seven-point scale to Asian and Latino party identification because of the high numbers of nonpartisans, especially among first-generation Latino immigrants, who are the focus of this present research. The authors argue that the scale and its assumption of linearity and a linear ideological spectrum do not explain the role of Independents well because “not all groups are equally familiar with the terrain of American politics” (2006, p. 134).

Independent Variable: Ideology of Latin American Governments

I use PELA data to measure Latin American government ideologies. This data offers a way to measure the ideology of Latin American governments on a 10-point left-right scale. PELA asked legislators, “On a scale where 1 is left and 10 right, where would you place the following political leaders?”—including the president. Using this scale, Alcántara-Saez (2008) classified political parties, leaders, and presidents from left to right. I used PELA to create a continuous measure that ranks the presidents from 1 (left-wing) to 10 (right-wing). The countries included in the analysis are Argentina, Bolivia, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Peru, Uruguay, and Venezuela. Although PELA has data about Paraguayan presidents, CMPS does not include Paraguayan respondents who arrived in the United States during the years to be analyzed. Cubans are the second largest Latino immigrant group in the United States, but PELA does not have data from Cuba. The Appendix contains the breakdown of Latino party identification by country of origin.

Left-right scales to measure ideology have been widely applied to parties and party systems in Latin America (Colomer and Escatel 2005; Coppedge 1997). Coppedge (1997), who analyzed ideology in 11 Latin American countries, recognized that the left-right dimension can be problematic for classifying Latin American parties but justified its use because, among other reasons, “it is usually clear to country specialists whether these parties are basically left of center or right” (p. 11)

Control Variables

I employed a set of demographic control variables derived from the literature. Prior research has demonstrated that Latino party identification varies by gender, education, and income (Alvarez and Bedolla 2003; Dutwin et al. 2005; Kosmin and Keysar 1995; Uhlaner and Garcia 1998; Wong 2000). Therefore, the models include a variable for gender and a six-point scaled variable for education. To control for income, I added a 12-point scaled variable for total combined household income. In a theoretical and historical discussion, some authors have argued that economically advantaged Americans are more likely to be Republicans, while the financially disadvantaged are more likely to be Democrats (Belknap and Campbell 1951; McCarty, Poole, and Rosenthal 2003). Considering the non-random distribution of ideology among countries, I added a control for Mexican origin, the largest country of origin in the sample. Mexicans are 53.5% of the respondents in the sample of Model 1.

All Latino respondents completed CMPS either in English or in Spanish. Considering the influence of language preference on Latino partisanship, I added a control variable for whether the respondent answered the survey in Spanish. To control for a possible religious influence, I added a control variable for self-identification as Catholic and another for self-identification as Protestant. I added a control variable for linked fate. I constructed a variable for linked fate with two questions from CMPS: First, respondents were asked, “Do you think what happens generally to Latino people in this country will have something to do with what happens in your life?” CMPS asked a follow-up question to those who answered “Yes”: “Will it affect you a lot, some, not very much?” I used these three options to create a 4-point variable for linked fate.

It is possible that the younger immigrants were when they migrated from their home countries, the fewer memories they have from their home countries, including about politics, especially if they migrated very young. Similarly, those who migrated later in their lives may have more explicit memories of their home countries. To control for such an effect, I added a variable for immigrants’ age on first arrival in the United States.

Results

Presidents with an ideological score closer to 1 are left-wing, and closer to 10 are right-wing. Using this scale, Alcántara-Saez (2008) used 4.99 as the midpoint in order to classify presidents as left- or right-wing. However, I applied the mathematical midpoint between 1 and 10 to classify presidents. To calculate the mathematical midpoint between 1 and 10, I use the Mathematical Middle Point formula:

$${{\left( {{x^1} + {x^2}} \right)} \over 2} = {\rm{Mathematical}}\;{\rm{Middle}}\;{\rm{Point}}$$

where $${x^1}$$ and $${x^2}\;$$ represent the endpoints of a line segment. The endpoints of the variable for the ideological score of Latin American presidents are 1 and 10. Applying the Mathematical Middle Point formula, the mathematical middle point between 1 and 10 is 5.50.

$${{\left( {1 + 10} \right)} \over 2} = 5.50$$

There are 14 left-wing and 39 right-wing Latin American presidents in the sample, for a total of 53. While all countries have had right-wing presidents, not all have had left-wing heads of state. Argentina, Bolivia, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Honduras, Nicaragua, Peru, and Venezuela have had both left- and right-wing leaders. By contrast, the Dominican Republic, Guatemala, Mexico, Panama, and Uruguay have only had right-wing presidents. Most left-wing leaders have been in power in the twenty-first century, following a recent leftist wave in Latin America.

Only five heads of state from the twentieth century were classified as leftists: Rodrigo Borja Cevallos (Ecuador), Daniel Ortega (Nicaragua), Ernesto Samper Pizarro (Colombia), Carlos Roberto Reina (Honduras), and Eduardo Frei Ruiz-Tagle (Chile). In the Appendix, I present the ideology index for each Latin American president in Model 1 by country and years in office, and the respective ideology coefficients calculated with PELA.

Latino immigrants in Model 1 are distributed by party identification: 52% identified as Democrats, 14.77% as Republicans, and 33.23% as Independents and “other party.” This distribution is similar to the distribution in CMPS, where 50.72% identified as Democrats, 14.95% as Republicans, and 34.33% as Independents and “other party.” The variable’s mean for government ideology in country of origin is 6.74. For more descriptive statistics from Model 1, see Table A.3 in the Appendix.

To determine the role of each independent variable in the relative probability that a Latino immigrant self-identifies as an Independent, Democrat, or Republican, I estimated a multinomial logit regression using STATA 16.1. Multinomial logit models require assuming the independence of irrelevant alternatives (IIA). Therefore, I performed a Hausman test to ensure that the assumption of IIA is satisfied (Hausman and McFadden 1984). The results of the Hausman test showed no evidence that the IIA assumption was violated.

The results of the analysis correspond to Model 1 in Table 2. The base category is Democratic self-identification. Since multinomial logit models produce nonlinear estimates, they can sometimes be challenging to interpret. To better understand the model, I reported the exponentiated coefficient estimates, best known as odds ratios or relative risk ratios, in addition to the multinomial logit coefficients. The odds ratio is a measure of the ratio of the probability of self-identification as either Independent or Republican to the probability of self-identifying with the reference party. For example, a coefficient of two on the Catholic variable for Republican self-identification (assuming statistical significance) means that if all other things are equal, a Catholic is two times more likely than a non-Catholic to identify as Republican. Odds ratios less than one indicate a negative association if statistical significance is present (Başlevent, Kirmanoğlu, and Şenatalar 2009).

Table 2. Latino Party Identification Model 1 (Multinomial logit model)

Notes: Standard errors are in parentheses. Data are weighted.

Government Ideology: higher scores mean that the government is more right-wing.

***p < 0.001; **p < 0.01; *p < 0.05; ⁺ p < 0.10.

The first pair of columns in Model 1 in Table 2 show the results for self-identifying Independent versus self-identifying Democratic. The government’s ideology in the immigrant’s country of origin at the time of migration is negative and statistically significant. Latinos who migrated from a country with a right-wing government are less likely to be Independents than to be Democrats. In the Republican versus Democratic identification results, the ideology of the government in the immigrant’s country of origin continues to be negative and statistically significant: Even when social and economic controls are in place, Latinos who migrated from a country with a right-wing government are less likely to be Republicans than to be Democrats. These results suggest that the government’s ideology in the country of origin seems to have an important effect on Latino party identification.

It is not surprising that linked fate is statistically significant. Latinos with a linked fate sentiment are less likely to be either Republican or Independent than to be Democrat. In addition, the language that respondents used in the survey is negative and significant for the probability of self-identification as Republican versus Democrat, but not for Independent versus Democrat. People who completed the survey in Spanish are less likely to be Republicans than Democrats. Having completed the survey in Spanish is negatively associated with the Republican Party. These results suggest that Latino immigrants who prefer speaking Spanish are less likely to self-identify as Republicans than to self-identify as Democrats, and that assimilated Latinos are more likely to be Republicans than Democrats. Indeed, another variable related to assimilation—age at migration to the United States—is statistically significant in the probability of being Republican compared to Democrat. These results suggest that Latino assimilation favors the Republican Party over the Democratic Party.

Income, gender, religion (both Catholic and Protestant), and education are not statistically significant. These results are similar to the findings of other studies in which income did not significantly affect Latino party identification (Alvarez and Bedolla 2003). However, findings from income and gender correspond to those of Alvarez and Bedolla (2003) regarding direction. By contrast, the results for income and education correspond to findings from Evans and colleagues (2012). Alvarez and Bedolla’s (2003) results for gender contradict Evans and colleagues (2012), but the outcome for gender in the current study corresponds with the findings of Alvarez and Bedolla. It is difficult to compare results regarding education with those of Alvarez and Bedolla because they coded education as four different variables according to the level of education, and there is no clear pattern among education levels.

It is important to analyze the size of effects and the predicted probabilities of government ideology in explaining party identification. Figure 1 shows the adjusted predicted probabilities for the likelihood of self-identifying as a Republican, as a Democrat, or as an Independent or “other party,” according to the political ideology of a Latin American immigrant’s country of origin at the time they migrated. Results are based on Model 1 in Table 2. Latino immigrants who identify as Republicans are more likely to have migrated from countries with left-wing governments, while those who identify as Democrats are more likely to have migrated from countries with right-wing governments. Although not part of the hypothesis, Latino immigrants are more likely to identify as Independents if they migrated from countries with left-wing governments.

Figure 1. Adjusted predicted probability of party identification with 95% Confidence Intervals

Overall, I find strong evidence supporting the hypothesis presented in this paper. Latin American immigrants are more likely to identify as Republican if they migrated from a country with a left-wing government, or as Democrat if they come from a country with a right-wing government. Further, I find even more robust evidence that Latino immigrants who migrated to the United States at an older age and those who completed the survey in Spanish are less likely to identify as Republican.

Robustness Checks

Latino experiences and marginalization are associated with political constructs (Sanchez and Masuoka 2010). Seven regression models were fit to explore the role that policy positions and attitudes have on the main associations. Each model accounts for individual’s priorities concerning the most important issue that the President and Congress should address. These issues are (1) wages, incomes, and minimum wage, (2) immigration reform, (3) health care, (4) racism and race relations (5) climate change, global warming, and environment, or (6) abortion and women’s reproductive health. The seventh model controls for the language in which Latino respondents consume news, in a 5-point scale ranging from mostly English-language to mostly Spanish-language. Tables A.5–A.11 in the Appendix report the results from these analyses. The associations derived from these models are consistent with those found in the main analyses.

Discussion and Conclusion

This research expands our understanding of acquiring party identity among Latino immigrants in the United States, emphasizing premigration experiences related to government ideology. Many scholars have studied Latino party identification, and the present research adds to this conversation by addressing an important gap. The analysis presented here suggests that premigration politics in the home country influences the partisanship of Latino immigrants. More specifically, these findings suggest that ideological differences in governments in the home country at migration can shape the partisanship of foreign-born Latinos in the United States. My analysis provides statistical evidence that the ideology of governments in the home country when Latinos migrated to the United States is associated with party identification at the individual level. The statistical analysis in this article shows a robust and statistically significant association between the dependent and independent variables of interest, and therefore support for the hypothesis.

I found a clear association between the language respondents used to complete the survey and party identification. Given that a preference for Spanish language over English is considered an indicator of an immigrant’s level of assimilation, this finding could suggest that a lack of assimilation into the United States is associated with a low probability of self-identifying as a Republican. The results of this research open up possibilities for new research on party identification targeting to immigrants from specific political backgrounds, and they represent a challenge to the applicability of classical partisanship theories to immigrants and Latinos.

Two possible theories explain the phenomenon observed in this study. The first is that individuals in their home country become disenchanted with their economic situation, so they decide to leave, and understanding the ideology of the current government in their country, they are more likely to affiliate with the U.S. party on the opposite end of the ideological spectrum. A second possible theory is that ideology drives individuals to become unhappy with their economic or political prospects and decide to leave. Once in the United States, these pre-existing ideological commitments align with the U.S. party closest to their ideological predisposition. Future research should determine how these ideas predispose people to migrate, but this is outside the scope of this present study.